Volume 9, Issue 2, Summer and Autumn 2009, Page 1-242

Multiple Regression Model Selection by Information Criteria

Zakaria Y. AL-Jammal

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 1-12
DOI: 10.33899/iqjoss.2009.30581

In this paper, I considered the problem of selection a model from a collection of candidate models specified by multiple linear regression model. Two criteria are used and compared, they are Akaike information criterion ( ) and Bayesian information criterion ( ). A real data set is considered as an application case. I preferred the model that the chosen rather than depending on the values of and the .

Bayes Estimator of one parameter Gamma distribution under Quadratic and LINEX Loss Function

Wael Abdul Lateef Jasim

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 13-28
DOI: 10.33899/iqjoss.2009.30586

In this paper we derive Bayes' estimator for the Scale parameter in Gamma distribution when is known and equal 2, i.e. , we take to estimate one parameter of gamma distribution which is (Scale parameter), where gamma distribution is considered as an important model of the life time models . These estimators are obtained depending on squared error and LINEX loss function , Then comparisons of risks for under squared and LINEX loss function have been made . Simulation study is given to illustrate that the proposed estimators is preferable to for the sample sizes from above distribution with parameters and for all values of " " .

A Suggested Algorithm for the Selection of Initial Centers of Clusters

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 15-28
DOI: 10.33899/iqjoss.2009.30531

In this research, we suggest an algorithm to select the initial centers of clusters; this algorithm includes the selection of features of different attributes as an initial center of each cluster. As a fruit of this algorithm it gives a unique cluster when processes any group of data many times. This result leads to save time and effort. Simulation experiments performed to test the proposed algorithm. Also, a real data deals with the food security of Arabic World is applied.

Fuzziness in linear programming with application

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 29-52
DOI: 10.33899/iqjoss.2009.30535

With regard to the difficulty and the big role that is made to reach the optimal solution of the institutions and factories system processing through the perfect or alternative decision making among the abundant decisions or alternative groups, and since that the information could be inaccurate or uncertain , in this research one of the types of non–linear logistic functions has been used, that is the modified s- curve membership function in the selection of the best production mixture for solving the problems of industrial institutions through the application of fuzzy linear programming method.
This function is qualified by including an important factor(fuzzy factor ) which could determine the shape of function as well as by its flexibility in dealing with fuzzy indications .
The industrial production units face the problem of being fuzzy in their various areas such as raw materials , human resources , work hours , …,etc. In order to solve this problem ,fuzzy programming method has been applied in this research in “the General Company for Vegetarian Oil ” to determine the best mixture and to achieve the required object increasing the profitability according to two important factor . the first factor is the level of satisfaction by taking (21) level which ranged between (0.0010-0.999) with an excess (0.0499) . The second one is fuzzy factor that ranged between (1-40) with an excess (2) that each one of them has been determined by the researcher .
The research has concluded that the optimal decision depends on the fuzzy Factor in the problem of determining the production mixture in the fuzzy model , as well as that the highest level production units could be obtained when the fuzziness in the model is low

Estimation of Weibull Distribution Parameters

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 53-72
DOI: 10.33899/iqjoss.2009.30540

This paper is concerned with the estimation of the parameters of Weibull distribution whose density function is given by

Samples for each of the different sample sizes were simulated 1000 times computer simulation is used to obtain the statistical properties of the estimators .
Estimators have been obtained by method of least squares,for the two cases, presence and absence of the second order approximation in the Taylor expansion, and comparison between the two cases is made. The estimators is less biased for the case where second approximation in Taylor expansion is present than for other case.
In order check the model examination of the residual is carried out for a sample of size 50 with 5 simulations for both in case of presence and absence of second order approximation in the Taylor expansion.
The errors have zero mean and constant variance.
The calculation is carried out by computer.

Estimation of Parameter for Gibbs Random Distribution which is used in Image Processing

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 73-94
DOI: 10.33899/iqjoss.2009.30543

In this research, the parameter was estimated for Gibbs random distribution using different energy functions. We get the first algorithm by adding noise to the image and restore it by filters. We obtain the second algorithm by degradation of the image and restore it to the original image or close to it. The third algorithm has energy function which depends on the mean and standard deviation.

Using The Probabilistic Fractal Dimension to Decide if a Time Series is Chaotic

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 95-114
DOI: 10.33899/iqjoss.2009.30548

In this paper, we use several ways to distinguish between the deterministic time series and stochastic white noise by using correlation exponent test and the BDS test ( Brock Deckert Scheinkman ) ,through simulating a large number of data for testing the efficiency of these methods on several different models.
We found that the correlation exponent tests can distinguish white noise from chaos, however it can not distinguish a white noise from a chaotic process mixed with small a mount of white noise .
By putting a white noise as the null hypothesis the BDS test rejects the null hypothesis when the data came from from stochastic source.

Pattern Matching Fuzzy Models for Time series Forecasting with Application

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 115-130
DOI: 10.33899/iqjoss.2009.30551

We study a new technique of forecasting time series , this is technique is pattern matching Fuzzy, which identify past relationships and its trends in historical data for forecasting future values .We connect Fuzzy Pattern Matching model with time series and we also make algorithm of Fuzzy Pattern Matching and application it on the real data (consuming electric energy in Ninevah Governerate) , and we used Mean Squared Deviation (MSD) and Mean Absolute Deviation (MAD) to get the optimal values

The Use of Robust Criteria in Selecting Effective Variables in Linear Regression Model for Blood Sugar Analogy

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 131-148
DOI: 10.33899/iqjoss.2009.30554

The research deals with the topic of variable selection in linear regression using robust procedures as resistant to outliers and other failures of assumptions . The objective of the research is using robust criteria (RAIC , RSIC, RCp, RVC, RSC, RF, RAPE) in selection of most adequate independent variables in the regression model used to estimate blood sugar as dependent variable and other independent variables and comparing the performance of these criteria. The results shows that the RAPE criterion was the best in selecting the most important variable compared with the other criteria.

An Algorithm Proposed to Find an Optimal Solution in Nonlinear Integer Programming

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 149-160
DOI: 10.33899/iqjoss.2009.30558

This paper proposed a new algorithm for solving a nonlinear integer programming for human resources allocation problem to find an optimal solution. It has been concluded that this algorithm is relatively simple and efficient for solving this type of problems, if it is compared to the other traditional approaches such as Kuhn-Tucker conditions and Lagrange multipliers method.

The Application of Fuzzy Logic to the Modeling of product Density for Children Ready-Made Clothes

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 161-184
DOI: 10.33899/iqjoss.2009.30561

The main objective of this research is to design a program model for a new product density estimation by implementing fuzzy logic techniques. This model is designed depending upon some of the factors influencing product density. The model consists of conditional rules. Mamdani fuzzy inference system is used for reasoning process because it is an efficient type of fuzzy inference for knowledge to make decision processing. The model is designed using MATLAB as the programming tool for writing the model's programs. The model is applied to real data average taken from Mosul factory for children Ready-Made clothes. The results obtained proved that FLPDEM is an attractive model for new product density.

Specification of the Conditional Expectation by Simple Linear Regression Model For Binomial Distribution Conditioned with Varying Sample Size.

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 185-202
DOI: 10.33899/iqjoss.2009.30564

In this research, we consider the study of conditional expectation and it's relationship with regression model. The conditional expectation has a linear form which is specified as a simple linear regression model. The power transformation was used on the predictor variable which gave the best possible fit for the model which was derived from the binomial distribution conditioned with varying sampl size.
The parameters of specified models were estimated by depending on emprical data which were simulated with different values for the parameter of conditional probability distribution. The best estimator for the power parameter was found in two specified models by the maximum liklihood and Draper & Smith methods. These estimators gave the best fit to the suggested model and best estimator to the conditional expectations of conditional probability distribution and it was concluded that the suggested method was better than the ordinary method.
The increments in the probability of success (p) had a great effect on the best fitted model also the estimated conditional expectation of conditional binomial distribution was affected. This result was clear because of decreasing the coefficient of determination (R2) in Draper & Smith and the mean square of residuals in maximum liklihood method with increase in (p).

Employment Bootstrapping Approach to Finding New Ratio Estimators in Simple Random Sampling

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 203-218
DOI: 10.33899/iqjoss.2009.30570

This research boils down to find new ratio estimators instead of estimators of (Kadilar and Cingi ;2004) by replacing the regression parameter of the final estimators, which is estimated by ordinary least squares, with new parameter estimated by bootstrapping regression under specified conditions which have more accuracy than the first estimators. The mean square error (MSE) was used to check the accuracy of new estimators, Then we fiend Relative Efficiencies for all proposed estimators. This work was supported with numerical examples and simulations .

Modified Robust M-approach to estimate the parameters of linear Regression model

IRAQI JOURNAL OF STATISTICAL SCIENCES, 2009, Volume 9, Issue 2, Pages 219-242
DOI: 10.33899/iqjoss.2009.30574

The idea of this research is to find construction Weighted Robust for the estimate parameters of the linear regression model by blend Robust M- approach and Weighted Least Squares method to handle the influence of Outlier Values, which don't harmonies with the system of data set, in order to get new estimators, The mean square error criterion is used to check the efficiency of the new method with OLS , WLS and Robust M-estimators.